A Circadian Rhythms Learning Network for Resisting Cognitive Periodic Noises of Time-Varying Dynamic System and Applications to Robots

• Published in: IEEE transactions on cognitive and developmental systems Vol. 12; no. 3; pp. 575 - 587
• Main Authors: Zhang, Zhijun, Deng, Xianzhi, Kong, Lingdong, Li, Shuai
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.09.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Biological neural networks; Circadian rhythm; Circadian rhythms; convergence; Dynamic models; Dynamical systems; Error functions; large errors; Learning; Mathematical analysis; Mathematical model; Matrix methods; neural network; Neural networks; Noise reduction; online equation solving; time-varying problem; Time-varying systems
• ISSN: 2379-8920; 2379-8939
• DOI: 10.1109/TCDS.2019.2948066

Time-varying dynamic system contaminated by cognitive noises is universal in the fields of engineering and science. In this article, a circadian rhythms learning network (CRLN) is proposed and investigated for disposing the noise disturbed time-varying dynamic system. To do so, a vector-error function is first defined. Second, a neural dynamic model is formulated. Third, a co-state matrix is integrated into the model, of which the states are the linear combination of the previous periodic states and errors, which can effectively suppress periodic noises. Theoretical analysis and mathematical derivation prove the global exponential convergence performance of the proposed CRLN model. Finally, a practical noise disturbed time-varying dynamic system example with four different noises illustrates the accuracy and efficacy of the proposed CRLN model. Comparisons with traditional zeroing neural network further verify the advantages of the proposed CRLN model.